Intelligent Automation: Eight Applications to Consider in 2020
Find opportunities to employ intelligent automation across the enterprise with these real-world use cases.
- By James Lawson
- February 25, 2020
As businesses plan for 2020, intelligent automation has become a top priority for many executives. Many enterprises have invested in robotic process automation (RPA) over the last few years; Forrester predicts the industry will continue to grow from $250 million in 2016 to $12 billion in 2023. RPA democratized coding, making it easy to automate tasks that follow simple rules and have structured data. Having made positive progress, executives are now looking to take the next step to further transform their operations.
According to a Deloitte survey, only eight percent of organizations have achieved "scale" with their RPA programs, defined as deploying more than 50 automations. Moreover, the limitations of RPA mean that most of these automations have minimal impact. RPA cannot handle the complexity of real-world businesses, which involve subjective decision making, unstructured data, and human interactions.
Intelligent automation holds much promise to overcome these challenges. Executives hope that technologies with greater intelligence will enable them to automate more broadly, deeply, and quickly. Despite much enthusiasm -- with 30 percent investing more than $50 million in their automation initiatives -- many don't understand what intelligent automation actually entails or how to achieve success.
Definitions and Early Applications
McKinsey aptly defined intelligent automation as "an emerging set of new technologies that combines fundamental process redesign with robotic process automation and machine learning." In short, add machine learning and redesign your processes. This allows businesses to automate work once deemed exclusively the domain of knowledge workers.
The first application of intelligent automation that often comes to mind is automating processes with unstructured data; for example, processing invoices. This is where early intelligent automation projects focused. Optical character recognition (OCR) tools are used to digitize documents, then machine learning is used to extract key fields that previously had to be read by people. There are many success stories here; the analyst firm Everest Group identifies this as its own market: intelligent document processing.
Intelligent document processing has delivered massive value in paperwork-intensive areas such as KYC (know your customer) verification and insurance claims. The main reason for caution is that this process typically struggles if the documents are low quality (e.g., low resolution) or too complex (length, language, structure, etc.).
Unstructured data was one of RPA's three core areas of weakness but focusing on data extraction misses the much larger prize that intelligent automation offers through data interpretation. Machine learning can then also be used to enhance decision making and human interactions.
Using Intelligent Automation in the Real World
There are arguably many more (and higher value) applications of intelligent automation when businesses reimagine their end-to-end processes. RPA acts as a gateway to machine learning-powered AI solutions. The software is used to gather data from different sources (particularly when APIs and integrations aren't available) and orchestrates responses based on the machine learning predictions, decisions, and recommendations (particularly quickly building reinvented process flows).
Eight cross-industry applications of intelligent automation have emerged with proven use cases for each:
1. Interpreting and processing unstructured data: Machine learning can be used to interpret and route customer queries, email, and tickets. Typically, they're classified into different categories based on type and priority. This then triggers RPA to orchestrate an automated or streamlined response.
2. Predictive analytics: Following a change in behavior or significant interaction, machine learning can make predictions about a customer, such as whether they are a high churn risk. RPA feeds AI with new customer data and orchestrates retention responses.
3. Predictive maintenance: Manufacturing and logistics industries rely on complex machinery. Analyzing time-sensitive data and sensor diagnostics with machine learning can detect fault risks. RPA is used to gather data from dispersed systems. Once a risk is identified, RPA can then manage the alert and schedule a proactive repair.
4. Radical personalization: Making bespoke offers for individual customers based on their history is a strategic imperative in consumer-facing and retail-based industries. RPA can gather customer data from across business lines and databases. Machine learning generates personalized recommendations, with RPA automating the delivery of these offers.
5. Trend/anomaly detection: The most common application of anomaly detection is identifying fraud. RPA automatically feeds new transactions while machine learning is used to identify outliers worthy of review. RPA then orchestrates the response, such as gathering more data and triggering an investigation.
6. Strategic optimization: Planning store locations is difficult to optimize. RPA can gather data from diverse and local sources and then machine learning can generate recommendations. Automation allows for richer and more current planning.
7. Forecasting: Forecasting is a common challenge across industries, from estimating demand for products in retail to managing nurse overtime in healthcare. RPA can be used to gather real-time and local data, and then time-sensitive machine learning can be applied to identify unexpected bottlenecks and improve accuracy.
8. Real-time optimization: In logistics, it's essential to optimize routing and identify issues proactively. Machine learning can enable preventative action for problem cases, with RPA providing real-time alerts to relevant teams.
These are just a few of the tried and true intelligent automation applications that enterprises should consider when planning for 2020. They push the boundaries of process transformation beyond discussions that focus only on RPA's unstructured data limitations. There are so many more opportunities to use intelligence to achieve greater return on investment from automation, with faster time to value.
About the Author
James Lawson is the AI evangelist at DataRobot. He is responsible for educating the market about artificial intelligence, accelerating adoption, and dispassionately advising executives about how best to achieve value from their transformation initiatives. Before DataRobot, Lawson was WorkFusion’s global head of strategic markets, a leader in RPA. He is a fellow of the Adam Smith Institute and read philosophy, politics, and economics at the University of Oxford.